A worksheet section is titled Percentage.
The learner sees “35%” and knows roughly what to do.
Another section is titled Speed.
The learner sees kilometres and hours and reaches for distance = speed × time.
Now remove the headings.
Mix ratio, percentage, area, average and a changing quantity inside one problem.
Can the learner still decide what to do first?
Transfer begins where the topic label ends. A learner owns a method when the mathematical structure can be recognised without being named in advance.
This is the purpose of a Primary 6 mixed-topic transfer diagnostic.
It is not another long practice paper.
It is a deliberately designed set of problems that changes the surface while preserving known mathematics, so that weak links in method selection become visible before examination conditions amplify them.
The quick answer: test the choice before the calculation
For each problem, record four decisions before evaluating arithmetic accuracy:
- Quantity identification: what is actually known and unknown?
- Representation choice: would a bar model, diagram, table, equation, number line or state chain reveal the relationship best?
- Method selection: which mathematical relationship should run first?
- Verification: what independent check could reject an impossible result?
A learner can be procedurally accurate and still fail at the second or third decision.
Why keyword hunting stops working
Words such as “altogether”, “remaining”, “of”, “per” and “difference” can provide clues.
They cannot determine the full model safely.
“Of” may signal multiplication in one percentage statement.
In another sentence it may simply be ordinary English.
“Difference” can describe subtraction, a comparison-bar gap, or a before-and-after change.
Primary 6 learners need a stronger habit:
identify the quantities and relationship before choosing the operation.
Diagnostic Layer 1: change the representation
Give a ratio problem first with a bar model.
Then give a mathematically equivalent problem using only words.
Then give the same relationship in a table.
If performance drops sharply after the visual support disappears, the learner may recognise the model without being able to generate it.
This is different from not understanding ratio at all.
Diagnostic Layer 2: move the unknown
Known form:
35% of 240 = ?
Transfer form:
84 is what percentage of 240?
Reverse form:
84 is 35% of what number?
The same relationship is rearranged.
If only the first form is reliable, the learner may know a procedure without understanding the roles of part, percentage and whole.
Diagnostic Layer 3: change the surface story
Use the same proportion in different settings:
- paint mixture;
- class composition;
- map scale;
- recipe;
- ticket allocation.
The arithmetic structure can remain identical while the nouns change completely.
If performance depends on familiar story templates, the concept has not yet generalised fully.
Diagnostic Layer 4: combine two familiar topics
A rectangular park measures 50 m by 30 m.
Forty percent is used for sports.
The remaining area is divided between gardens and paths in ratio 3:2.
Find the garden area.
Dependency chain:
- park area = 50×30 = 1,500 m²;
- remaining percentage = 60%;
- remaining area = 900 m²;
- ratio total = 5 units;
- garden = 3/5 of 900 = 540 m².
No individual step is exotic.
The transfer demand is knowing that the ratio applies to the remainder, not to the original area.
Diagnostic Layer 5: combine three topics
A circular track has circumference 440 m.
A runner completes 5 laps in 20 minutes.
Find average speed in m/min.
Distance:
440×5 = 2,200 m.
Average speed:
2,200÷20 = 110 m/min.
The learner must connect circle measurement, repeated multiplication and rate.
Diagnostic Layer 6: change units halfway
A tank contains 2.4 m³ of water.
Water is drained at 80 L/min.
How many minutes are needed to empty the tank?
2.4 m³ = 2,400 L.
Time:
2,400÷80 = 30 min.
A learner who starts with 2.4÷80 has a unit-compatibility failure, not merely a division error.
Units are part of method selection. If the quantities cannot interact dimensionally, the calculation should not begin.
Diagnostic Layer 7: remove the obvious diagram
Word problems often become easier after a good representation is built.
The diagnostic question is whether the learner can build that representation independently.
Give a before-and-after ratio problem with no bars.
Ask the learner to decide whether a bar model, unit table or equation would be most useful.
Score the representation choice separately from the final answer.
Diagnostic Layer 8: require estimation before exact work
Before calculating 38% of 1,980, ask for a rough range.
40% of 2,000 is about 800.
So the exact answer should be near 750–800.
If the learner later obtains 75.24, the magnitude check should reject it immediately.
Transfer requires not only producing answers but monitoring whether they make sense.
Diagnostic Layer 9: ask for two valid methods
Find 25% of 96.
- fraction route: 1/4×96 = 24;
- decimal route: 0.25×96 = 24;
- percentage-unit route: 10%=9.6, 20%=19.2, 5%=4.8, total 24.
Then ask which method is most efficient and why.
Comparing correct methods reveals whether the learner understands structure beyond one algorithm.
Diagnostic Layer 10: include irrelevant information
Real problems do not always present only the numbers needed.
Add one irrelevant detail and ask the learner to explain why it is unnecessary.
This tests quantity selection rather than calculation.
Using every number because it appears in the question is not mathematical reasoning.
Diagnostic Layer 11: use a familiar method in an unfamiliar direction
A cuboid has volume 360 cm³ and base area 45 cm².
Find height.
height = volume ÷ base area = 8 cm.
A learner who only remembers l×w×h may hesitate when the formula must be used backwards.
This is inverse-transfer testing.
Diagnostic Layer 12: delayed return
Do not test transfer only immediately after teaching.
Return several days later with a changed surface and reduced cueing.
Immediate success can reflect short-term familiarity.
Delayed, changed-condition success is stronger evidence that the relationship has become retrievable and portable.
A 45-minute mixed-topic diagnostic
- one ratio/percentage representation switch;
- one rate problem with unit conversion;
- one composite geometry problem;
- one circle or area problem with a remainder;
- one volume problem requiring reverse use of a formula;
- one data interpretation problem;
- one before-and-after problem;
- one fully integrated unfamiliar problem.
The exact number of questions matters less than the variation in mathematical demand.
Record the first broken decision
Useful diagnostic notes sound like:
“Converted ratio to fraction correctly after a bar model was drawn, but did not generate the model independently.”
“Correct speed formula recalled; failed to convert minutes to hours before substitution.”
“Calculated 30% accurately but applied it to the original total rather than the remainder.”
These observations tell a tutor what to repair.
“Weak at word problems” does not.
Common misconception 1: unfamiliar story means unfamiliar mathematics
Many unfamiliar problems are familiar relationships wearing new nouns.
Repair: strip away names and identify quantities, units and relationships.
Common misconception 2: the first possible operation is the first correct step
A number can be usable in several arithmetic operations but belong to only one meaningful relationship.
Repair: name the intermediate quantity before calculating it.
Common misconception 3: more working means better reasoning
Long working can hide a wrong model.
Judge whether each line produces a quantity needed by the next line.
Common misconception 4: correct answer proves transfer
A familiar cue may have triggered the procedure.
Repair: change representation, move the unknown and return later.
Common misconception 5: every difficult question needs a clever trick
Many difficult questions become ordinary after the dependency chain is exposed.
Complexity can be organisational rather than conceptual.
A transfer diagnostic ladder
- Can the learner solve a routine version?
- Can the learner solve it without a topic heading?
- Can the learner generate the representation independently?
- Can the learner solve after the unknown moves?
- Can the learner solve after units change?
- Can the learner solve after the surface story changes?
- Can the learner combine the idea with another known topic?
- Can the learner estimate before exact work?
- Can the learner explain why the selected method fits?
- Can the learner verify by a second route?
- Can the learner return days later and still reconstruct the method?
How this fits Primary 6 and PSLE preparation
The updated October 2025 MOE Primary Mathematics syllabus organises content across Number and Algebra, Measurement and Geometry, and Statistics, with problem solving at the centre of learning. PSLE Mathematics then samples that knowledge through multiple-choice, short-answer and structured/long-answer items.
This diagnostic is not an official PSLE instrument. Its job is narrower: test whether the learner can select and connect known mathematics when the question no longer announces the topic.
The deeper lesson: method selection is itself mathematical knowledge
Knowing how to multiply fractions is knowledge.
Knowing when a problem requires fraction multiplication is also knowledge.
Knowing when another representation would be clearer is a further layer.
The final stage of Primary Mathematics is not simply knowing more methods. It is recognising which relationship owns the next step when no keyword tells you.